Jakarta, Indonesia · Open to Remote
I secure AI systems before they break.
Indonesia's AI security researcher specializing in federated learning security and fraud detection systems. I find vulnerabilities in ML pipelines and build the defenses to stop them.
EXPERTISE
Security research grounded in real-world production experience.
Core Research Focus
Attacking federated systems to find weaknesses, then building the defenses. My work spans Byzantine attacks, model poisoning, backdoor detection, and cryptographic trust verification.
Industry Experience
Production-grade fraud detection for Bank Rakyat Indonesia. I've maintained 99.9%+ SLA on real-time transaction monitoring systems handling millions of events.
PROVEN RESULTS
My most impactful work in AI security and fraud detection.
Novel defense combining ECDSA signatures with anomaly detection and reputation scoring to protect federated learning from poisoning attacks.
Comprehensive research implementations covering the full FL threat landscape: Byzantine-robust aggregation (Krum, FoolsGold), poisoning attacks (label flipping, backdoor, model), privacy techniques (DP-SGD, secure aggregation), and gradient leakage analysis.
Static analysis framework for banking trojan detection using PE file analysis, API call sequence analysis, YARA rules, and ML classification.
CONTACT
Building FL defenses? Scaling fraud detection? Seeking a research collaborator or MPhil candidate? Let's talk. Based in Jakarta, open to remote and research positions.